90% and Seventeen Basis Points (11 September 2026): August's Inflation Prints Repriced the Fed in Three Sessions — $100 Oil Had Moved the 2-Year Two

August core CPI came in at 0.3% on the month, a tenth above forecast, and the odds of a quarter-point increase at next week's Federal Reserve meeting went to 90% on Friday from about 72% on Thursday and 59% three sessions earlier. The 2-year Treasury note rose seventeen basis points across the same window. The same note had moved two basis points in the sessions when Brent crude crossed $100. The market did not reprice the Fed on the oil shock. It repriced on two backward-looking domestic price reports that were collected before the oil shock happened — and that is the most useful thing this fortnight has to teach.

This page was published on 23 August as a preview of the Jackson Hole keynote and has tracked the September decision since. Its finding on 9 September was that a crude spike through $100 had barely moved the pricing of the meeting, and that the two inflation reports still to come were both August data — pre-shock by construction. Both have now landed. They were indeed pre-shock, and they were more than enough.

What actually happened: the prints did in one session what the oil did in none

The cleanest way to see the fortnight is to put the energy move and the rates move in one table, because the contrast is the finding rather than the background.

Yields through 10 September are the official daily par yield curve published by the US Treasury; spreads are calculated from those figures. *Friday levels are intraday quotes reported by CNBC during the session, not official closes. Probabilities are CME fed funds futures and the FedWatch tool as reported by CNBC.

Read down the 2-year column. From 4 to 8 September — the window in which Brent added roughly four dollars and crossed a round number that led every market report on two continents — it moved two basis points. From 8 to 10 September it moved seventeen, and thirteen of those landed in the single session after the producer price report. The 5-year did eighteen, the 7-year sixteen, the 10-year fifteen, the 30-year twelve. All three benchmark tenors closed at their highest levels of 2026.

Then Friday did something worth noticing on its own. The 2-year rose almost eleven basis points intraday on the CPI release and gave nearly all of it back, closing the reported session up 2.2bp at 4.572%. The 10-year actually finished lower, down 3.4bp at 4.91% after approaching 5.00%, and the 30-year fell 4.3bp to 5.318%. A market that has moved from 59% to 90% does not have much left to price for the meeting itself. What it has left to price is everything after it.

Why the oil did not do this and the CPI did

This is the mechanism worth keeping, because the intuition it corrects is very widely held and this fortnight is an unusually clean test of it.

A central bank targets a rate of change in a broad price index over time. An oil price move is a change in one relative price. When crude rises from $96 to $100, the energy component of a price index is higher for as long as the level is higher — and then, twelve months later, the same level contributes nothing to the annual rate, because the comparison base has caught up. A permanent step in the level of oil produces a temporary bump in the rate of inflation. That is the arithmetic reason a policy rate does not have to respond, and it is why central banks have spent forty years talking about core measures that strip energy out.

What converts an energy shock into a monetary problem is second-round effects, and there are two channels to watch rather than assume:

The August core CPI print is a reading on the third box, not the first. It measures what happened to the price of goods and services with food and energy removed, in a month that ended before the escalation. A tenth of a percentage point above forecast on that series is worth more to a rate-setter than four dollars on a barrel, because it is evidence about the trend rather than about a level. That is the whole reason seventeen basis points followed two prints and two basis points followed a crude spike, and it is not a quirk of this particular week — it is the reaction function working exactly as written.

The producer price report makes the same point from the other direction, and it is the detail most of Thursday's coverage flattened. Headline final-demand producer prices accelerated hard, to 5.44% over twelve months from 4.84% in July. But the measure that strips food, energy and trade services — the one built precisely to show the trend rather than the shock — eased, to 4.66% from 4.73%. The wholesale gate is where an energy shock shows up first, and it duly showed up. Underneath it, nothing accelerated.

Waller's condition, and what the data did to it

Governor Christopher Waller framed the vote on 3 September by attaching a date to it rather than a direction, which is why his words are the right yardstick for what just happened.

He conceded that inflation was "meaningfully above" the 2 percent target but said recent trends "suggest we are finally seeing some signs of disinflation", then made the conditional explicit: "If this continues in the data due over the next two weeks, I would be inclined to support holding the target for the federal funds rate at its current setting." He put it more plainly a moment later — "I'm going to paraphrase John Lennon here: Give disinflation a chance. We can wait one meeting."

Most headlines dropped the symmetric half. "I judge that policy is currently only slightly restricting aggregate demand, and it may not take much acceleration in inflation to nudge me into supporting tighter policy," he said. "If there is evidence that progress toward 2% inflation reversed in August, a small adjustment in our stance would help ensure that it resumes."

The data due has now arrived, and it is genuinely mixed against that standard.

Actuals are computed from BLS index series; consensus figures are as reported by CNBC.

Three of those six lines are in line, one is a tenth hot, and two of the trend measures eased. It is not the reversal Waller named as his trigger, and it is not the continuation he named as his reason to wait. Chairman Kevin Warsh's breadth test — he found 54 percent of the 199 components of the PCE price index rising faster than 3 percent at Jackson Hole, against 32 percent in the two decades before the pandemic — is the more demanding standard, and the annual core readings easing on both the consumer and wholesale sides is the kind of evidence that weakens rather than strengthens his case. The market nonetheless moved to 90%. What that gap says is that the Committee's centre of gravity, on a 9-3 vote in July with three participants already preferring an increase, did not need much.

The long end had an official buyer and sold off anyway

The doubled Treasury buyback operations this page flagged on 9 September are now observable rather than scheduled, and the result is instructive.

On 19 August the Treasury announced it was increasing, "by at least double", the size of its liquidity support buybacks for longer-dated nominal coupon securities, raising the maximum from $2 billion per operation to at least $4 billion, effective 9 September and running to 4 November. On Thursday 10 September it bought back about $5.2 billion of off-the-run 10-year notes and 20-year bonds — roughly half of the $10.5 billion dealers offered, per CNBC. The 10-year yield rose 11 basis points that day regardless, touching 4.954%, its highest since October 2023.

That is the useful separation. A $5.2 billion official bid is a supply-side intervention at the back end; an eleven-basis-point selloff on a producer price report is a repricing of the expected path of the policy rate. They act on different parts of a yield, and the second is larger this fortnight. Note also what the curve did while both were happening: 2s30s flattened from 86 to 81 basis points between 8 and 10 September, because the front end rose seventeen and the back end twelve. A buyback that presses the long end and a print that lifts the short end produce the same shape from opposite causes.

What it means for the dollar

The interest-rate factor is one of the five the meter scores, and the last week is a clean demonstration of both halves of that qualification — one factor is not a forecast, and size still matters.

Through early September the dollar was weakening while US front-end yields inched higher, which looked like a paradox and was not. A dollar index is a relative price, and a large share of it is the yen, whose central bank meets on 17-18 September with a 25 basis point increase to 1.25% widely expected — the day after the Fed announces. A two-basis-point improvement in US rate expectations cannot outweigh a repricing that size on the other side of the cross.

Seventeen basis points is a different quantity, and the direction changed with it.

European Central Bank daily reference rates. The ECB's own decision on 10 September is the other half of the euro leg.

The commodity factor still cuts against the intuitive direction: the United States is a net energy exporter, so a crude spike is not the terms-of-trade shock for the dollar that it is for Japan or the euro area. The fuller decomposition of the energy move, and why destroyed hulls have not yet been lost barrels, sits on the Hormuz page.

What would change the picture

At 90% priced, the decision itself is close to a non-event in the arithmetic sense: a quarter-point increase on Wednesday removes very little uncertainty, because very little is left. Three things around it are not priced.

The dot plot. September is one of the four meetings carrying a Summary of Economic Projections, and only two meetings follow it — 27-28 October and 8-9 December, per the Federal Reserve's published calendar. Bloomberg reported that bond traders moved to price two increases by the end of the year after the CPI report, which is a claim about those two dates rather than about this one. A hike with a flat set of dots and a hike with a second increase written into them are different events behind the same headline, and the distinction will matter more to the curve than the decision. That it comes from a chairman who spent his tenure arguing against routine forward guidance is the standing irony of this meeting.

The oil that is not in the data. September CPI, due mid-October, is the first official index that can contain the escalation. If the pass-through channel is real it shows up in core goods with a lag of months, and it shows up in refined products first — where the diesel crack spread set an all-time record in August with crude far below its 2022 level.

The Bank of Japan, the day after. The yen leg of the dollar index reprices on 17-18 September on a different calendar entirely. A Fed that hikes into a BoJ that hikes is a smaller dollar event than a Fed that hikes into a BoJ that holds.

The five-factor framework exists so that a read does not have to collapse into one number, and this fortnight is the case for it. The commodity factor moved a great deal and moved the policy expectation almost not at all. The rate factor moved little and then moved a lot, and only the second move was large enough to turn the currency. Same weeks, same market, three different channels — and the one that repriced the Fed was the quietest of them.

Educational macro context only — not investment advice.